Canonical URLs: handle duplicates without losing useful pages
Understand canonical, redirect and noindex; review tracking, filters, hostnames and language variants before changing signals.
Updated: · AI-generated editorial illustration
A canonical URL is a proposed representative for duplicate or very similar content. It helps clarify which version should be preferred, but it is not a guarantee that Google will choose the declared address. Contradictory sitemap, internal-link and canonical signals create additional interpretation work.
This guide covers common business situations: www and non-www hostnames, campaign parameters, catalog filters and bilingual articles. Avoid a universal remove-all-queries rule because some parameters genuinely select a different language or content set.
1. Distinguish tools with different purposes
Canonical proposes a representative while the current page can remain accessible. A redirect moves visitors and systems to another URL. Noindex asks for exclusion from results through supported mechanisms, with access needed for the instruction to be discovered. These tools are not interchangeable merely because a plugin presents them together.
Google's canonical documentation explains methods and signals. A permanently moved address often needs a redirect for the user journey rather than an indefinitely surviving old page with a canonical hint. A distinct page intended for indexing should not casually point at another page.
2. Map the sources of duplicate versions
Inspect hostnames, HTTP versus HTTPS, trailing slashes, tracking queries, filters, sorting and print views. Record the URL, whether content changes, response state and current canonical. Similar-looking addresses alone are not enough; examine the content and task. A meaningful filter destination is not automatically a duplicate of its parent category.
For tracking-only UTM parameters, content usually stays the same and representative signals should remain coherent. A language parameter selecting Vietnamese or English changes the content and needs suitable addresses and alternates. Google's multilingual-site guidance supports checking version discovery rather than treating translation as duplication.
3. Align canonical, sitemap and internal links
Choose a representative that is accessible, serves the intended content and is not accidentally noindexed. Use a clear address on the primary domain. Internal links should target the representative where appropriate, and the sitemap should list intended discovery URLs. A www sitemap and canonical pointing to a retired demo domain are contradictory signals.
Social metadata, structured data and mainEntityOfPage need the same context. After changing domain configuration, inspect public HTML rather than source alone. An old hosting environment value can preserve an incorrect canonical even after a code change. Verification belongs on the rendered deployment as well as in the repository.
4. Evaluate category and filter pages by value
For a large catalog, distinguish filters helping visitors find products from destinations with their own indexable value. Avoid generating thousands of indexable combinations without useful content. Equally, do not point every paginated page to page one without examining the content and product discovery path.
Google's ecommerce URL guidance supports this review. Sample categories, empty filters, populated filters and product pages before changing a rule. The category-page SEO guide helps make the content decision first, then select the technical instruction appropriate to that decision.
5. Inspect the representative Google actually selected
URL Inspection can show the declared canonical and Google's selected version where data is available. If they differ, examine duplication, redirects, sitemap, links and response states before concluding that the system is wrong. A desired separate page may lack sufficiently distinct content or consistent signals.
Record release and inspection dates. A live test does not mean the stored index has refreshed. Monitor after crawler processing instead of relying on a local screenshot. There is no responsible fixed-date promise that every URL will adopt your selected representative. Treat the inspection as evidence for the next investigation, not a complete explanation by itself.
6. Preserve URL history during consolidation
When consolidating articles, map old to new according to equivalent answers. Update content links, sitemap and breadcrumbs. Avoid redirect chains and blanket homepage destinations, which usually do not answer the retired article's question.
Check old and new URLs after release, including status, content, canonical, alternates and image access. Keep the mapping for rollback or existing external links. Read the migration and redirect guide before broad URL changes. A single canonical edit is not a substitute for a migration plan covering the visitor journey and the site's historical addresses.
Frequently asked questions
Should two articles on one topic share a canonical?
Not automatically. Distinct questions and content can have their own representatives. When intent and content overlap, make an editorial consolidation decision before changing technical signals.
References and further reading
To apply this to your current website, explore the SEO optimisation service or contact Dean Vu to define a scope around your existing site and actual evidence.